Development of a pattern recognition system for discriminating osteoarthritic bone marrow edema like lesions on MRI

Georgios Katsiberis; Pantelis Georgiadis; Aspasia Rigopoulou; Dimitrios Siamplis; Elias Panagiotopoulos; Ekaterini Solomou

Volume : IV, Issue : IX, September - 2015

Development of a pattern recognition system for discriminating osteoarthritic bone marrow edema like lesions on MRI

Georgios Katsiberis, Pantelis Georgiadis, Aspasia Rigopoulou, Dimitrios Siamplis, Elias Panagiotopoulos, Ekaterini Solomou

Abstract :

A non-invasive method developed to investigate the capacity of digital image texture analysis for the detection and differentiation of hip ostaoarthritic lesions. Thirty one femoral heads, after hip arthroplasty, were examined on a 1.5T MRI system. Bone marrow lesions (edema and cysts) were graded on MRI. Utilizing these MRI series, volumes of interest were extracted and classified for every ostaoarthritic lesion. According to the developed pattern recognition system, the highest classification accuracy in discriminating bone marrow edema from bone cysts was 98,92%, employing five textural features (mean value, correlation, sum of squares, sum average and sum variance). Individual accuracies in classifying correctly bone marrow edema lesions were 98, 73% and bone cysts 99, 07%. Using the developed pattern recognition system, accurately grading the severity of hip osteoarthritis can be performed, discriminating reversible and irreversible lesions in order to monitor the progression of the disease and the therapy effectiveness.